4. REGRESSION MODELS
Shyam-Sunder and Myers (1999) present two simple models to assess to what extent a firm’s financing behavior can be explained by the pecking order- or the trade-off theory. Also, we present a third model which includes conventional leverage factors following Frank and Goyal (2003). All variables are defined in the Data Appendix.
4.1 The Pecking Order Model
The pecking order assumes no target level of debt, rather the capital structure is a product of the firm choosing capital according to the preference order; (1) internally generated funds, (2) debt, and (3) equity. The model suggested by Shyam-Sunder and Myers (1999) states that when a firm's internal cash flows are inadequate for its real investments and dividend commitments, the firm issue debt.
To test this, we examine financing decisions made after short-term changes in profits and investments, by using the theoretical relationship between changes in the level of debt and a firm's need for funds. The theory states that the level of debt issued or retired from the company should be adjusted according to the firm’s financial needs when taking all variables that form the earlier financing deficit as exogenous. By doing so, the level of debt increases or decreases depending on whether or not the requirements of the investments can be covered by the internal cash flow.
10
We test the pecking order hypothesis with the following model:
∆𝐷𝑒𝑏𝑡 = 𝐷'.) - 𝐷'.)+,= a + 𝑏-./𝐷𝑒𝑓'.) + 𝑒'.)
Where 𝐷 is the long-term debt to assets and 𝐷𝑒𝑓'.) is the funds flow deficit . Equity issues and repurchases are not included in the financial deficit as the theory predicts that a firm will only issue or retire equity as a last resort. The pecking order hypothesis is that the financing deficit is completely covered by debt. That is, a = 0 and 𝑏-./ = 1. The financial deficit is being calculated as:
𝐷𝑒𝑓'.) = 𝐷𝑖𝑣'.) + 𝐼'.)+ ∆𝑊𝐶 '.) - 𝐶𝐹'.)
Where Div:.; is the cash dividends calculated as the change in dividends payable plus the dividend expense, I:.; is the capital expenditures calculated as the change in fixed assets minus depreciation, ∆WC :.; is the working capital calculated as current assets minus current liabilities and CF:.; is the cash flow.
As described by Shyam-Sunder and Myers (1999), the sign of the deficit is irrelevant in the simple pecking order model. If a company has a surplus (the deficit being negative), and the only imperfection is information asymmetry, all managers will end up paying down debt. If there are tax or other costs of holding excess funds or paying them out as cash dividends, the managers will have a motive to repurchase shares or pay down own debt. Managers who are less optimistic than investors will pay down debt, instead of repurchasing shares at a too high price. More optimistic managers will try to repurchase own shares but will force stock prices up. As the price increases the number of optimistic managers decreases, which in turn leads to even higher stock prices. As a result, if information asymmetry is the only imperfection, the repurchase price is so high that all managers end up paying down own debt (Shyam-Sunder & Myers,1999).
4.2 The Trade-Off Model
The static trade-off theory predicts that firms aim to keep a constant target debt ratio. When firms experience a deviation from the target, they respond by increasing or decreasing their capital to reach the optimum. We test the trade-off theory with the following model:
∆𝐷𝑒𝑏𝑡 = 𝐷'.) - 𝐷'.)+,= a + 𝑏/./ (𝐷 ∗'.)+𝐷'.)+,)+ 𝑒'.)
Where 𝐷 is the amount of debt issued or retired and 𝐷 ∗'.) is the target debt level at time t.
We test the hypothesis that 𝑏/./ = 1, which implies that the debt level equals the target level (𝐷'.)=𝐷 ∗'.)). However, it is likely that transaction costs will occur when adjusting towards the target. Therefore, it is reasonable to assume that firms will allow a certain deviation from the target level before adjusting. A 𝑏/./ between zero and one indicates an adjustment towards the target while 𝑏/./ above one implies an over-adjustment.
As the target debt level is unobservable for companies, 𝐷 ∗ has been derived by taking the average of the historical mean debt ratio for each firm and the historical industry mean. Following Shyam-Sunder and Myers (1999) we also test using the historical mean of the debt ratio for each firm, not including the industry average.
4.3 Pecking Order Model with leverage factors
When trying to explain the level of a firm’s leverage, a common approach is to test the relative importance of the factors which empirically is said to influence a firm’s financing decisions.
Following Frank and Goyal (2003), we have run a regression model with the leverage factors in first differences. First differences may bias the leverage factors towards zero. However, this approach makes it appropriate to nest the financing deficit variable into the equation, to see the explanatory importance of this variable.
The model contains four leverage factors which should affect the level of leverage according to the pecking order theory. Further, the financing deficit is an added
12
To conduct this study, we need accounting information on Norwegian firms. The data is obtained from the database of the Center of Governance Research (CCGR), which contains accounting information on all private and non-private Norwegian firms. The relevant data is available from the year 2000 to 2015.
In 2006, there was a reform in the Norwegian taxation law of dividends. The reform incentivized firms to increase their leverage ratios up to the year 2005 before sharply reducing them (Alstadsæter & Fjærli, 2009). Therefore, we select 2006 as the starting year for our analysis.
Firms are included in the final sample if they have a minimum of five years of data on the relevant variables. Previous tests of the trade-off model eliminate firms without continuous data (Jalilvand & Harris, 1984; Titman & Wessels, 1988).
Testing for only the pecking order theory does not require continuous data.
To reach our final sample, several filters are applied. First, financial firms are excluded as they face specific regulations regarding capital structure. Second, daughter firms are excluded as they have a capital structure decided by their parent company. Third, firms with zero revenues or zero employees are removed as we define these firms as non-operating. Further, firms with inconsistent accounting information such as negative debt, depreciation or fixed assets are excluded. In addition, negative equity firms are removed from the data as these firms may distort the results. The final sample contains 63,503 unique firms, resulting in a data panel with 412,474 observations.